According to the WHO, unsafe sex is responsible for 1.7 million deaths in low income countries and is the leading risk factor for mortality in African women. An increasing number of sexual behavior surveys are conducted in sub-Saharan countries to understand and address the key determinants of poor sexual/reproductive health, but such surveys present major limitations. They are often affected by non-negligible levels of non-response and only provide self-reported, highly unreliable data on individual behaviors. Virtually all surveys have also been egocentric, i.e., they are limited to eliciting the sexual partnerships of a respondent without providing information on the sexual networks of a respondent's partner(s) that indirectly put the respondent at risk. These indirect connections are a major determinant of disease dynamics in populations. In 2005, we initiated a unique study of population-level sexual networks on Likoma Island (Malawi) that integrated partner tracing. Using these data (also known as sociocentric), we were able to reconstruct maps of the sexual connections that link inhabitants of the island directly and indirectly. We have also documented large biases in self-reported data on sexual partnerships. In 2007, we conducted a follow-up of that study, yielding the first longitudinal dataset on the global sexual networks of a sub-Saharan population. The main aim of this application is to document previously unobserved longitudinal dynamics of the sexual networks that make sex unsafe in Likoma (a small island of Northern Malawi). During this project, we will 1) use multiple reports of sexual partnerships obtained during the survey to measure the extent of non-response, attrition and misreporting biases in self-reported sexual network data, 2) describe, for the first time in a sub-Saharan setting, changes in the global structure of sexual networks and changes and individual trajectories within these networks and 3) determine whether sexual and reproductive health outcomes are associated with the observed network structures. This will provide new information on sexual networking that could inform preventive interventions in sub-Saharan countries. It will also provide empirical estimates of parameters that will permit the appropriate calibration of mathematical models of disease spread. PUBLIC HEALTH RELEVANCE: We use a unique dataset collected in Likoma (Malawi) and in which sexual partnerships were traced to (a) investigate patterns of reporting biases in sexual behaviors by comparing sociocentric and egocentric information on sexual behaviors, and (b) document the longitudinal dynamics of sexual networks on Likoma to better understand how sexual behaviors create risks for sexual and reproductive health. This will provide new information on sexual networking that could inform preventive interventions against sexually transmitted diseases in sub-Saharan countries.